RE: Problem Found! Need solution

[Scott Klement <klemscot@xxxxxxxxxxxx>]

On Tue, 17 Sep 2002, Hopkins, Bill wrote:
>
> It's the fact that 4B is NOT equivalent to BINARY(4) that catches most folks
> when they're first doing this.  Maybe Jon can shed a little light on that
> particular mystery.
>


It shouldn't take an expert like Jon to explain why BINARY(4) is different
from 4B...  it has, after all, been discussed here *MANY* times before.

Remember this:

       1) The API manual is not intended to be specifically for RPG
           programmers.  It is intended for anyone writing software
           in any language on the iSeries.

       2) Every different language has a different way of referring to
            a 32-bit binary integer.   In C, it's called "int", but
            on some other platforms (such as MS-DOS or 16-bit Windows), an
            int might be only 16 bits.   In CL, you'd use a CHAR(4) but
            put %bin() around it.  In RPG IV, its correctly defined as 10I.
            In RPG III you had to define it in a data structure, by
            putting a "B" in column 43, and using from & to locations
            of 1-4 or 5-8, etc, etc, etc

       3) So, to try to make the meaning clear, the API manual refers
            to the number of bytes long the variable needs to be.
            4 bytes is 32 bits.  That should be very clear.   Now you
            just have to be familiar with your programming language to
            know how to define it.

       4) In RPG, numeric variables (with the exception of floating point
            which was added later) are not defined by the number of bytes
            that they take up, but rather are defined by the number of
            DIGITS that they can hold.   i.e. a  "13P 3" is a 13 digit
            packed variable, it is only 7 bytes long.   Likewise, a
            "4B 0" variable can hold up to 4 digits, but is only 2 bytes.

Then, to add to the confusion, there's the differrence between the
"9B 0" and "10I 0" data types.

You see, a 32-bit binary integer can store values from -2147483648 to
2147483647.  Even though those many seem like completely random numbers
in decimal, in binary they are nice neat numbers, the highest values
possible in 32 binary digits.

But the nasty "9B 0" pseudo-binary number only stores values from
-999999999 to 999999999, even tho they are considered "binary" they get
cut off on a decimal digit boundary.  The APIs aren't expecting this.

Thats why I always tell people to use "10I 0" instead of "9B 0".  It
correctly uses all possible values of a 32-bit integer.

What they should do is remove the "B" from the RPG manual, and have a
compiler message that says "The B data type is deprecated, please
replace it with a I" just like they do in every other language on every
other platform.

I know, I know, you can use decimal places with a "B" data type and not
with an "I", and people think that this means "B" should stay.   But,
really, who uses that?!